Jackknife with user written cluster program

[Felix Bittmann]

Dear all,
I would like to use the jackknife with a user written program and use clustering (paneldata). However, my program fails and I am not sure why. With a bootstrap this works fine. Can someone point out my folly?

clear all
webuse pig
gen newid = id
xtset newid week

cap program drop xtest
program define xtest, rclass
    xtreg weight c.week
    return scalar coef = r(table)[1,1]
end

*Test program
xtest
return list

*Jackknife
jackknife r(coef), cluster(id) idcluster(newid): xtest

The output of the program:

. jackknife r(coef), cluster(id) idcluster(newid): xtest
(running xtest on estimation sample)

Jackknife replications (48)
1 ---+--- 2 ---+--- 3 ---+--- 4 ---+--- 5
nnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnn
insufficient observations to compute jackknife standard errors
no results will be saved
r(2000);

[Fei Wang]

Felix, jackknife needs the user-written program to allow for the if qualifier.

program define xtest, rclass
    syntax [if]
    xtreg weight c.week `if'
    return scalar coef = r(table)[1,1]
end

[Felix Bittmann]

Thanks, much appreciated!